LAUSR

Considering the great importance of flood prediction, flood routing based on Shark Algorithm (SA) and Four-Parameter Nonlinear Muskingum (FPNM) has been proposed in the present study. In fact, the Muskingum model is considered as one the most efficient method for predicting flood. However, to successfully implement Muskingum Model, there is a need to compute various parameters of this model utilizing a lot of data that characterized the physical features of the catchment. Therefore, there is a need to integrate the Muskingum model with an optimization method. Nevertheless, there are several drawbacks including trapping in local optima, overhead response and convergence time-consuming have been experienced using the existing optimization methods. Therefore, in this study, a proposal for utilizing an integrated evolutionary computing method namely; SA with FPNM has been introduced to overcome such drawbacks. Three case studies based on the definition of objective functions and different error indices were used to evaluate the algorithm. The results showed that SA significantly reduced the sum of the total square deviations (SSQs) and the sum of absolute deviation (SAD) between the predicted and observed discharges compared to other evolutionary algorithms. Moreover, the proposed model achieved high ability to accurately determine the peak value and peak time of the discharge. In addition, the calculated hydrodynamic shape has a high correlation with observed hydrographs.